[Brinkman]

I've been reading a bit about the strengths and weaknesses of the SRPP topology, which is the topology employed in Jim's Clarinet linestage. Despite the author's best intentions, it's still not very easy for me to wrap my head around all of it. Nonetheless, my reading has brought to mind a couple questions. Here's the papers to which I refer to, written by Merlin Blencowe.

From what I can gather, the load is a crucial element of the SRPP circuit's performance. If you know the load of the SRPP circuit (i.e. the input impedance of the device(s) plugged into the output of the Clarinet), you can optimize the resistance values in the circuit for that known load. I'm not sure what optimized load for the Clarinet in stock form might be, but I plan on running my Clarinet into two devices simultaneously: a power amp (100K ohm input impedance) and an active subwoofer (12K ohm input impedance). I have no clue whether such a load is too small for the Clarinet (which increases distortion), a bit higher than the target load (not bad, but defeats the purpose of an SRPP), or right on target (optimal).

For that reason, I dare suggest something like a table of useful resistor values would be helpful to the end user, allowing them to further optimize their Clarinet for desired loads.

Additionally, there's a newer modification on SRPP, called SRPP+, that allows one to equalize current flow between the triodes. It consists of an additional resistor between the cathode of the top triode and the output tap. Information about that is included in the above linked paper.

I understand there was already a Hagerman-approved Clarinet modification from a few years back:

Remove the four 680 ohm resistors at R 308 and R310.

Replace the R 308 resistors with 1.2k ohm resistors and the R310 gets 820 ohm resistors.

It struck me that this modification was more about dropping the current draw of the SRPP circuit. However, in light of some of the reading I mentioned above, I am curious if there is more to it than that.

Included below is the clarinet schematic for reference:


Best regards.

[poty]

Hello,
I say sorry in advance, because several my understandings are not follow the ideas of the mentioned articles.
Several years ago I came to the article and found it useful as far as the results were clearly explained and very close to the existing circuits' values. But... Then I found a lot of line stages which greatly contradict with the findings of the article. It amused me to dig deeper. In the process of reading the article I discovered several doubtful points and contradicting assumptions, but mainly the article was innocent in its description. I tried to count some values for line stages according to the article and found that they are completely unusable. The puzzle has a very simple answer.
The article named "The Optimised SRPP Amp", so the values which are counted in the article are suitable for some narrow margins of usage, particularly - for using the SRPP circuit for "This article will look ... how it
can be optimised for its task as an output-transformerless power amplifier." In the line stage there are several differences in the goal and optimization, amplitude of the signal (both input and output) and so on. But why the article is so appealing to the readers? I found a key phrase in the article which is both marvelous in its simplicity and not true in its nature. "for proper balance the quiescent current must be half the peak current value" which, as author wrote, automatically "optimizes distortions". I doubt this circle is suitable for theory discussions, entering off-topic area, so I'll not write problems which are sourced from this assimption.
Answering your question: with its output impedance of 3k - all loads from probably 10k will be suitable.

[cheap-Jack]

Hi.

(1) the weaknesses of the SRPP topology,

(2) I plan on running my Clarinet into two devices simultaneously: a power amp (100K ohm input impedance) and an active subwoofer (12K ohm input impedance).

(1) The "weakness" of SRPP in general is it is NOT all cure though the topology does attracts tons of DIYers out there to try building it. It is designed to drive certain range of load. Otherwise it will not be better than any basic cathode followers.

If the gain of yr driving preamp is high enough, why NEED to go thru the hazzle of building a SRPP. Wouldn't a simple 12AU7 cathode follower do a good joB.

Don't forget to have 2 separate heater circuits for the upper half tube & for low half tube as the cathode voltages of both upper & low halves are different by a huge margin, risking blowing the tubes if no such precation is taken.

(2) Taking the rules-of-thumb, the ideal I/P impedance should be min 10X, so O/P Z of the SEPP should be 12/10=1.2KR. But the SRPP get 3KR O/P Z. A bit too high for yr sub.

Mind you, it is the sub that may cause audibly O/P signal voltage drop because of its low low I/P Z of its built-in SS amp.

I am running into such similar O/P voltage drop problem with my active sub hooked up direct to my passive linestage & I am building an very simple impedance buffer for my sub, hoping to fix such issue.

c-J

[poty]

(1) The "weakness" of SRPP in general is it is NOT all cure though the topology does attracts tons of DIYers out there to try building it. It is designed to drive certain range of load. Otherwise it will not be better than any basic cathode followers.

Yes, this thoughts could be easily picked from the article, but some practice tell us some opposite things.

If the gain of yr driving preamp is high enough, why NEED to go thru the hazzle of building a SRPP. Wouldn't a simple 12AU7 cathode follower do a good joB.

Too many "if"s... SRPP has its own beautiful things, which can considerable complicate "a simple cathode follower" to the extent that it became a nightmare to build (aside of expensiveness) in comparison with the SRPP.

Don't forget to have 2 separate heater circuits for the upper half tube & for low half tube as the cathode voltages of both upper & low halves are different by a huge margin, risking blowing the tubes if no such precation is taken.

While many DIYers do that - personally I do not follow the precautions, mainly because do not use weak tubes. Having 200V specs on cathode-heater pair, getting 280V for both (140 for a half) is very easy goal.

(2) Taking the rules-of-thumb, the ideal I/P impedance should be min 10X, so O/P Z of the SEPP should be 12/10=1.2KR. But the SRPP get 3KR O/P Z. A bit too high for yr sub.

Without conditions - your rules-of-thumb nothing more than stereotype. For example: If you read the article you can remember the main thought - the effectiveness could be achieved if the load is equal to the output impedance.

[Brinkman]

Don't forget to have 2 separate heater circuits for the upper half tube & for low half tube as the cathode voltages of both upper & low halves are different by a huge margin, risking blowing the tubes if no such precaution is taken.

Are you referring to a THINGY, per Morgan Jones? Or two separate 6.3V supplies for each tube?

At any rate, I'm still doing my homework on the circuit. For instance, the updated value for Rak (1.2Kohm vs 680) lowers the output impedance and the max current swing into the next stage. My calculation - if I have my figures correct - is that the output impedance is now 2000K. Not sure how much the max current swing was reduced but I'll eventually figure it out.

Also considering increasing the output coupling cap to 4.7uF but I'm running out of room.

[Brinkman]

Okay.

So I've been reading some other technical chapters articles dealing with SRPP circuits. More elementary coverage was found in the third edition of Morgan Jones' Valve Amplifiers (pgs. 120-125). Also helpful was John Broskie's three-part series of TubeCAD blog entries on SRPP+ from 2009: Blog entries 0171, 0172 & 0173. Blog entry 0172 being extremely helpful in that it begins with a hyperlinked index of all his prior SRPP articles. Additionally, the instruction manual for his SRPP+ PCB kit can be read here. Page 17 of the aforementioned manual contains a table of  data (mu, r_p, R_k) for common 9-pin tubes at various B+ voltages. 

The following formula is Amos and Birkinshaw's formula for calculating the output impedance (Z_o) of an SRPP circuit. In these formulas, the subscript ₂ merely denotes the top triode and ₁ the bottom triode:
Z_o=[r_p2(R_k+r_p1)]/[r_p1+r_p2+R_k(mu₂+1)]

Using the Clarinet's stock values (R308 & R310 are 680 ohms):
Z_o=[8900(680+8900)]/[8900+8900+680(16.6+1)]=
85262000/29768=2864 (approx)
Don't forget to add the series output resistor R311 (220 ohm):
2864+220=3084

In the next formula, we can determine the low frequency shoulder (-3dB cutoff) of the Clarinet. Where C is the output capacitor's value in microFarads (uF) and R is the output impedance (Z_o) of the Clarinet:
Hz_-3dB=159155/C/R

Using the stock values (1uF coupling cap and 3K output impedance):
159155/1/3000=53Hz [NOTE: The specifications for the Clarinet state a bandwidth of 10-100K, so I must be missing something here.]

Doubling the size of the coupling cap adds an extra octave of low frequency:
159155/2/3000=26.5Hz (doubling the cap again - to 4uF - would give you yet another octave).

The following formula is for determining the optimal impedance (Z_opt)for the SRPP circuit. In other words, the optimal input impedance that the Clarinet should see "downstream." Having an input impedance higher than the optimum value is fine, but much higher begins to negate the advantages of an SRPP topology. An input impedance lower than optimal is to be avoided:
Z_opt=(mu₂*R_k2 - r_p2)/2

Again, using the Clarinet's stock values:
Z_opt=(16.6*680-8900)/2=1194
Using the Clarinet's modified values (R308 is 1200 ohm):
(16.6*1200-8900)/2=5510

So raising R_k2 from 680 to 1200 raised the Z_opt from a measly 1194 to 5510. Additionally, as seen in the first equation, a higher R_k also lowers the Z_o with the tradeoff being a lowered maximum current swing into the load (see Broskie's articles).

Anyway, I have some more helpful equations and a few questions I need to ask but it's getting late so I'll stop here.

[poty]

In the next formula, we can determine the low frequency shoulder (-3dB cutoff) of the Clarinet. Where C is the output capacitor's value in microFarads (uF) and R is the output impedance (Z_o) of the Clarinet:
Hz_-3dB=159155/C/R

Using the stock values (1uF coupling cap and 3K output impedance):
159155/1/3000=53Hz [NOTE: The specifications for the Clarinet state a bandwidth of 10-100K, so I must be missing something here.]

The formula doesn' take into account the amount of Z_load. I've weighted the resulting formula up and it should be:
Hz_-3dB=708/(1+1834*C*(Z_o + Z_load))
(resistance in Ohms, capacitance in Farades)
If we speak about loads more than 30k the 10Hz became the reality.

The following formula is for determining the optimal impedance (Z_opt)for the SRPP circuit. In other words, the optimal input impedance that the Clarinet should see "downstream."

It's again a question: what the formula optimizes? At first glance it looks like the formula from Blencowe's article, so it is for optimal power, not voltage!
Also I doubt that we should drop the lower and higher cathode resistors difference in case of modified values and more than that - the presence of the anode resistors comparable in the value with the cathode resistors.

[Brinkman]

It's again a question: what the formula optimizes? At first glance it looks like the formula from Blencowe's article, so it is for optimal power, not voltage!

Okay, I gathered a bit of that from your earlier statement in regards to Blencowe. Also, Broskie hints at that when discussing current swing effects due to raising R_k. Perhaps these facts are of more importance when the goal is some sort of output transformerless application - such as a headphone amplifier.

I guess what I am most curious about is how Jim H went about the change from the stock R_k values to the modified R_k values. In the stock build, both cathode resistors are 680 ohm, which seems to match the description of the circuit better. It can be assumed that tests of this build proved favorable in Jim's setup. So what puzzles me most is not how raising the values of the cathode resistors led to an improvement (though I think that aspect is worth discussing), but how imbalancing them led to an improvement (i.e. 680 & 680 vs 1200 & 820).

Here's John Broskie discussing this topic. Here he refers to R_k2 as R_ak. He is discussing breaking up R_k2 into a voltage divider consisting of R1 and R2 with the output being taken from the join of R1 and R2:

In effect, the external load is mitigating the tube's transconductance, so the value of resistor R_ak (R2) must be increased by 2R_load/mu. In other words, a different load impedance, a different R_ak value. If the triode offered infinite mu, one Rak resistor value would work with all load impedances. This means that setting up an SRPP circuit takes much more thought than many solder slingers can marshal, but armed with a calculator and the formula finding the optimal resistor value is not difficult. The actual difficulty occurs when we try to use our correct value in an actual circuit. For example, a 6H30-based impedance-multiplier circuit working into a 32 ohm load requires an R_ak value of 73 ohms, which when used with a 100V differential from cathode to plate, will result in an idle current draw of over 40mA, which comes dangerously close to exceeding the tube's dissipation limit. In addition, we may only want to allow 20mA of idle current. The solution to this problem is the SRPP+, as its additional resistor allows us to set the idle current to 20mA and to retain the impedance-multiplier ratio of 2. Resistor Rak has been replaced by two resistors, R1 and R2, in the SRPP+ circuit. The first step is to set value of resistor R1 and R2 combined, as it must match that of the bottom triode's cathode resistor. For example, the 6H30 requires a cathode resistor of 220 to establish an idle current of 20mA with cathode-to-plate voltage of 100V. The next step is to use this value as Rk in the following formulas:

R2 = r_p/2mu + R_load/mu + R_k/2
R1 = R_k - R2


Still trying to wrap my head around all of this.

Also I doubt that we should drop the lower and higher cathode resistors difference in case of modified values and more than that - the presence of the anode resistors comparable in the value with the cathode resistors.

Poty, would you please restate this? I'm not sure I follow.

[poty]

Hi again,
My returning to tubed circuits was not far ago, so I have not tried everything to count and simulate yet, but I can guess several things which maybe give you a hint for understanding or ground for critic of my thoughts. I'd be glad if you point me for bad guess.

Okay, I gathered a bit of that from your earlier statement in regards to Blencowe. Also, Broskie hints at that when discussing current swing effects due to raising R_k. Perhaps these facts are of more importance when the goal is some sort of output transformerless application - such as a headphone amplifier.

Regarding the optimization: I see the efforts of the optimization (in the way it is discussed in the articles) as "maximization" of output power. Lets look at the ratings for Clarinet. Output voltage is 1Vrms, quiescent current (original) 6 mA. If we assume that this is the amplitude, then we have around 4 mArms maximum to the load (all assumptions I got from the articles). 1V/4mA=250 Ohm. So, the max output (minimal impedance) for which the original circuit is capable of is 250 Ohm. You can guess, that it is not realistic for a preamplifier, especially tubed, to be designed with such low-impedance load. Just to estimate... "Optimum" load should be equal to the output impedance of the circuit, which is 3k, so we have "headroom" here - about 12 times. It means that to drive optimum load we can afford the 12 times less current swing! It gives us more possibilities to tweak the values not for power gain or maximum current swing, but for less distortions and wider usage base.

I guess what I am most curious about is how Jim H went about the change from the stock R_k values to the modified R_k values. In the stock build, both cathode resistors are 680 ohm, which seems to match the description of the circuit better. It can be assumed that tests of this build proved favorable in Jim's setup. So what puzzles me most is not how raising the values of the cathode resistors led to an improvement (though I think that aspect is worth discussing), but how imbalancing them led to an improvement (i.e. 680 & 680 vs 1200 & 820).

Let's look at the John Broskie's article 173:
"the SRPP holds two sub-circuits and that one of the two, the impedance-multiplier circuit"
"the bottom triode’s impedance at its plate only influences the impedance-multiplier circuit’s output impedance, but has no effect on the R1 and R2 values, even if the bottom triode were replaced by a pentode. The bottom tube’s impedance at its plate will alter the output impedance seen by the load. The lower the impedance at the plate, the lower the output impedance."
I hope you catch the main thing? The upper and lower sections don't have to be equal! It is true if we have conditions to work in the A-region for both sections and don't exceed the datasheet limitations of the tubes. As far as we found ten-times "headroom" we can afford to tune impedance matching and output impedance separately. I can't be completely sure it is the case of the changing, but it seems that I'm not far from the initial intentions.

Here he refers to R_k2 as R_ak. He is discussing breaking up R_k2 into a voltage divider consisting of R1 and R2 with the output being taken from the join of R1 and R2:
Still trying to wrap my head around all of this.

The idea is to allow the SRPP to drive even less impedance than the one limited by tubes specs and corresponding R_ks.

Poty, would you please restate this? I'm not sure I follow.

Sorry, I'm not very good at English, so sometimes write not clear.
In the phrase I tried to point out two differences in the Jim's version comparing to "classical" version:
1. You had found that R_k1 not equal R_k2 in new Jim's version, but still used "symmetrical" formulas to determine the SRPP stage specs.
2. In the Jim's version there are two 220 Ohm resistors connected to plates. Top resistor resembles R_a from the "classic" version, but low triode resistor is absent from any articles description. Its value is only 3 times lower then R_k (in original version) and in my opinion must have some influence on the circuit spec.

[hagtech]

I wish I could remember why I came up with that resistor change - mostly I believe it was a re-biasing to lower the plate voltage on the bottom tube.  It may have been to bring the heater-cathode voltage of the upper tube to a more satisfactory level.  I remember some folks having trouble with certain 12AU7 tubes that would bias high and start to "squeal".  That's what you hear, sort of a whistle, when the insulation is starting to break down.  There may have been something else too, perhaps a shift in current or increasing input headroom, I really can't remember.  I did some work once measuring grid current versus operating point.  You can do that if you have a VacuTrace [and a modification to the grid stopper].  Maybe that also played a role.  I'm a bit irked with myself for not writing this down somewhere.

The change had nothing to do with optimum power transfer or anything like that.  Also, I am not familiar with the articles mentioned and have not gone through Broskie's work.  They sound like interesting reads.

jh

[hagtech]

Here are some plots I made 11 years ago testing a 12AX7.  The first shows grid current as a function of grid voltage at a fixed plate of 50V. 



The next is with a fixed grid bias of -0.5V and sweeping plate voltage.  Of course, plate current changes too...



The main thing you'll note is that you do not want to operate too close to zero grid bias voltage.  A good operating point takes into consideration quite a number of parameters.

jh

[Brinkman]

So I modified the Z_o formula to account for the 220 ohm plate load resistors (R_p) in the Clarinet. Basically, I can't see any reason they shouldn't be added in series with the dynamic value r_p:

Z_o=[(r_p2+R_p2)(R_k+r_p1+R_p1)]/[r_p1+R_p1+r_p2+R_p2+R_k(mu2+1)]

Using circuit values;
Z_o=[(8900+220)(680+8900+220)]/[8900+220+8900+220+680(16.6+1)]=2959

Adding the series output resistance (R311);
Z_o=2959+220=3179

Interestingly, subbing a 12BH7 (r_p=5660; mu=15.7) for the 12AU7 lowers the output impedance by about 40 percent;

Z_o=[(5660+220)(680+5660+220)]/[5660+220+5660+220+680(15.7+1)]=1670
Z_o=1670+220=1890




Anyway, I reread John Broskie's three blog entries dealing with his "SRPP+" and more and more sinks in with every reading. Posted below is the circuit design for his SRPP+ iteration. It differs from Jim's Clarinet design in that there are no plate load resistors and no grid stopper on the upper triode. A series output resistor is not specified.



The "+" in SRPP+ is meant to indicate an SRPP circuit substituting a voltage divider (R1 & R2 above) for R_k2. The voltage divider maintains the desired amount of resistance for a certain bias point of an SRPP circuit, while also tapping the circuit's output from a point where the current contribution between the tubes is better equalized, lowering distortion.

In Broskie's own words:

For example, the following formula tells us what the optimal load impedance is for a given triode and R_ak value:

     R_load = (mu*R_ak – r_p) / 2

Thus, if we use a 6H30 with 250V B+ power supply and 300-ohm cathode and R_ak resistors (mu = 15.4 and r_p = 1380), the optimal load impedance will be 1620 ohms, which is great if that is the load we wish to drive. But what if we planned on driving 300-ohm headphones instead? Then we would use the following formula which gives the optimal Rak value for a specified triode and load impedance:

    R_ak = (2R_load + r_p) / mu

   
I know I already brought up the following formulas, but I would like to run through it with numbers to show how Broskie's circuit changes work and why - as I have now realized - the "+" will not work with the Clarinet. The formula in the quote above provides the ideal R_k for a given load. In the provided example, the load is 300 ohm headphones:

     R_k = (2R_load + r_p) / mu[/quote]
      =[2(300)+1380]/15.4
     R_k = 129 ohms

The problem is that 129 ohms is so low that the triode dissipation limit is exceeded. Using the formula below, Broskie shows how the R_k can be converted into an optimized voltage divider that maintains the 300 ohm resistance that keeps the dissipation to a safe level and helps drive low loads:

    R2 = r_p2/2mu + R_load/mu + R_k2/2
    R1 = R_k1 – R2

Solving for R2;
    R2 = 1380/30.8 + 300/15.4 + 300/2
    R2 = 44.8 + 19.5 + 150
    R2 = 215 ohms

Solving for R1;
    R1 = R_k1 - R2
    R1 = 300 - 215 = 85 ohms

The problem for us is that the load in this scenario is 300 ohms. Using the values found in the stock Clarinet circuit (R_k2=680; R_p=8900; mu=16.6) and assuming the sort of load an power amplifier might present (50K), we run into a problem:

    R2 = r_p2/2mu + R_load/mu + R_k2/2
    R2 = 8900/33.2 + 50000/16.6 + 680/2
    R2 = 268 + 3012 + 340 = 3620

    R1 = R_k - R2
    R1 = 680 - 3620 = -2940

A negative number means the Clarinet cannot be optimized with those values. Using Jim's modified values (R_k2=1200; R_k1=820) doesn't help either:

    R2 = r_p2/2mu + R_load/mu + R_k2/2
    R2 = 8900/33.2 + 50000/16.6 + 1200/2
    R2 = 268 + 3012 + 600 = 3880

    R1 = R_k1 - R2
    R1 = 820 - 3880 = -3060

Subbing a 12BH7 (r_p= 5670; mu=15.7) for a 12AU7 just makes it more incompatible;
    R2 = 5670/31.4 + 50000/15.7 + 1200/2
    R2 = 181 + 3185 + 600 = 3966

    R1 = R_k1 - R2 = 820 - 3966= -3146

How about decreasing the load to 10K?
    R2 = 5670/31.4 + 10000/15.7 + 1200/2
    R2 = 181 + 637 + 600 = 1418

    R1 = R_k1 - R2 = 820 - 1418= -1876

So basically, the problem we encounter when trying to employ the SRPP+ for linestage use is that the sort of loads your average linestage is going to run into are too large for this sort of modification. In fact, using the 12BH7 figures from above, you would need to be running the Clarinet into 1K ohm loads or less in order to optimize it.

So my main concern is that while I'm fine running the Clarinet into the 100K ohm load my tube amp would present, running the Clarinet into that same amp and an active subwoofer simultaneously (12K ohm input impedance; combined load of about 10K) would have some drawbacks. Exhausting any sort of tweak to fine-tune the Clarinet circuit for this sort of duty, I'm probably best off making only one change; increasing the output capacitor to 2.0uF:

    Hz_-3dB=708/[(1+1834*C*(Z_o + Z_load)]
    =708/[1+1834*.000002*(1890+10000)]
    = 16
[an output capacitor of only 1.0uF with a 10K load yields a -3dB of only 32Hz, which is too high]

Thanks Poty!

[poty]

So I modified the Z_o formula to account for the 220 ohm plate load resistors (R_p) in the Clarinet. Basically, I can't see any reason they shouldn't be added in series with the dynamic value r_p:
Z_o=[(r_p2+R_p2)(R_k+r_p1+R_p1)]/[r_p1+R_p1+r_p2+R_p2+R_k(mu2+1)]

Did you use the formula for bypassed R_k1? The Clarinet is unbypassed.

Using circuit values;
Z_o=[(8900+220)(680+8900+220)]/[8900+220+8900+220+680(16.6+1)]=2959
Adding the series output resistance (R311);
Z_o=2959+220=3179

I forgot to mention earlier, but adding the series resistance - why do not you add the impedance of the output C303?

... the circuit design for his SRPP+ ... differs from Jim's Clarinet design ... there are no plate load resistors and no grid stopper on the upper triode. A series output resistor is not specified.
The "+" in SRPP+ is meant to indicate an SRPP circuit substituting a voltage divider (R1 & R2 above) for R_k2. The voltage divider maintains the desired amount of resistance for a certain bias point of an SRPP circuit, while also tapping the circuit's output from a point where the current contribution between the tubes is better equalized, lowering distortion.

You have described the differences right, but you should read the purpose of the "+" before sinking into it. As I understand the "+" - we should go this way ONLY if the classic SRPP can't do the duty for low-resistance load. You mentioned much more prominent load that even "optimum" for classic SRPP - why you should be bothered by SRPP+ at all? "+" here doesn't mean "better sound", "advanced", "better quality" at all! It means - suitable for certain conditions outside the normal SRPP usage! That is why (while we do not have such conditions in the preamplifier) the SRPP+ is not suitable for Clarinet.

So my main concern is that while I'm fine running the Clarinet into the 100K ohm load my tube amp would present, running the Clarinet into that same amp and an active subwoofer simultaneously (12K ohm input impedance; combined load of about 10K) would have some drawbacks. Exhausting any sort of tweak to fine-tune the Clarinet circuit for this sort of duty, I'm probably best off making only one change; increasing the output capacitor to 2.0uF:

    Hz_-3dB=708/[(1+1834*C*(Z_o + Z_load)]
    =708/[1+1834*.000002*(1890+10000)]
    = 16
[an output capacitor of only 1.0uF with a 10K load yields a -3dB of only 32Hz, which is too high]

OK! There is some truth in your words. I don't think that the circuit itself would have some drawbacks, but if you really want to be on the safe side you can increase the capacitor of course.
What is more: Depending on the sensitivity of your amplifier and subwoofer I'd add another R311, C303, R312 to the circuit for the second output and tweak R311 resistances (make them higher) to lower demands for C303s.

[Brinkman]

Did you use the formula for bypassed R_k1? The Clarinet is unbypassed.

The formula, as I originally saw it, was showing an unbypassed R_k.

I forgot to mention earlier, but adding the series resistance - why do not you add the impedance of the output C303?

Because one can increase C303 to 2.0uF without penalty (besides space and cost) and did not want to bother with more equations for something that is irrelevant above 1K.

You have described the differences right, but you should read the purpose of the "+" before sinking into it. As I understand the "+" - we should go this way ONLY if the classic SRPP can't do the duty for low-resistance load. You mentioned much more prominent load that even "optimum" for classic SRPP - why you should be bothered by SRPP+ at all? "+" here doesn't mean "better sound", "advanced", "better quality" at all! It means - suitable for certain conditions outside the normal SRPP usage! That is why (while we do not have such conditions in the preamplifier) the SRPP+ is not suitable for Clarinet.

Well, when I brought up the SRPP+ I was wondering if Jim's modifying of the R_k values was done for similar concerns. After more reading, I understand the "+" better, and can see that it is of no use for line stage loads. Beyond that, Jim more or less mentioned that he simply changed the R_k values to set a bias point less taxing on new production 12AU7s. Rather than just abandon the thought at that point (since I brought it up anyway), I figured it would be more instructive for anyone following along to see why it is of no use mathematically.

What is more: Depending on the sensitivity of your amplifier and subwoofer I'd add another R311, C303, R312 to the circuit for the second output and tweak R311 resistances (make them higher) to lower demands for C303s.

Not sure I understand the theory behind such a change. Would it be to tailor each output to the input sensitivity of each amplifier? Might not be a bad idea.


As far as the Clarinet's concerned, I have a couple NOS Amperex Bugle Boy 12BH7s that I intend on running in it. As far as I know, the 12BH7 is a drop-in replacement for the 12AU7, providing the H+ circuit can supply the extra current. As I learn more, I may find the circuit can be better tailored to the 12BH7 and may have a few changes to make along those lines.

Beyond all that, I have a soft-start delay module that will fit inside that I plan on using, as well as a solid-state rectifier substitute for the 5Y3GT. Additionally, I also have a pair of dual Voltage Regulators/Constant Current Sources that I plan on using between the power supply and R306.

[poty]

Not sure I understand the theory behind such a change. Would it be to tailor each output to the input sensitivity of each amplifier? Might not be a bad idea.

The theory is simple: you can find better resistor easier than better capacitor. And cheaper too.
Your power amplifier probably has input sensitivity 1-2V and input impedance in the margins of 50k-100k. It will work fine with current parts.
Your subwoofer (I presume - active) has input sensitivity in the range of 250-500mV and you mentioned 12k as input impedance.
The Clarinet has gain of 15dB (5.6 times), so a source should have not less than 2 / 5.6 = 360mV - most of the modern sources are able to output more than 1V, so we even have some spare power here.
So, the first idea - to normalize the sensitivity of the following inputs. If we use a 12k resistor in series with the subwoofer input we effectively lower the sensitivity of the subwoofer to 0.5-1V and raise the input impedance to 24k!
The second idea - to eliminate any capacitors doubling along the signal path. You should be able to determine if your power amplifier or your subwoofer have input capacitors. If yes - maybe it's better to drop the capacitor(s), preferably in the devices near the end of the chain (because the other way you have high voltage in the interconnect).

As far as the Clarinet's concerned, I have a couple NOS Amperex Bugle Boy 12BH7s that I intend on running in it. As far as I know, the 12BH7 is a drop-in replacement for the 12AU7, providing the H+ circuit can supply the extra current. As I learn more, I may find the circuit can be better tailored to the 12BH7 and may have a few changes to make along those lines.

I don't have Clarinet, but use the SRPP circuit as line stage in the integrated amplifier. I use Russian 6Н1П with success (like it more than several AU7-th I have on hand, have less output impedance). Another versions: ECC802, 6Н6П, 6H8C, 6SL7, 6DJ8/E88CC...

Beyond all that, I have a soft-start delay module that will fit inside that I plan on using, as well as a solid-state rectifier substitute for the 5Y3GT. Additionally, I also have a pair of dual Voltage Regulators/Constant Current Sources that I plan on using between the power supply and R306.

Well... first addition is more "service" than sound. Constant Current Sources? Where do you intent to put them?

[Brinkman]

If we use a 12k resistor in series with the subwoofer input we effectively lower the sensitivity of the subwoofer to 0.5-1V and raise the input impedance to 24k!

I think I follow you well enough. I still would like to raise the value of the output capacitors to at least 1.5uF, but I will also purchase a pair of 12K 1W resistors to complement the suggestion. As far as redundant capacitors, I'm not sure about my 500W subwoofer plate amplifier, but I know my tube amplifier has no input capacitors because it has input transformers!

Anyhow, most of my current reading on the circuit is in regards to finding the purpose of the 220 ohm plate load resistors in the Clarinet schematic. Curious that most SRPP designs omit these resistors.

Well... first addition is more "service" than sound. Constant Current Sources? Where do you intent to put them?

Well, I need to have some sort of B+ delay to replace the soft start I lose by replacing the 5Y3GT with a solid state rectifier. Additionally, the slow ramp up of voltage should make things easier on the cathodes of the preamplifier tubes. The time delay relay / soft start module is powered by the 5V rectifier heater tap, but is essentially out of the circuit after the B+ has finished it's "ramp up." Not needing to power the heater of a tube rectifier should also free up some transformer flux for the extra heater current draw of the 12BH7s (vs. 12AU7s).

The high voltage shunt regulators / constant current sources are the second group of items seen here. One for each channel will go between C301 and R306 in the Clarinet schematic. Coincidentally, if you visit the link and click on "Power Amplifier" on the left menu, you will see the same tube amplifier I am now using and plan to use in conjunction with the Clarinet.

BTW, your list of Russian tubes was very interesting. I was particularly curious about employing a pair of 6Н6П tubes in a circuit like the Clarinet...

[poty]

Anyhow, most of my current reading on the circuit is in regards to finding the purpose of the 220 ohm plate load resistors in the Clarinet schematic. Curious that most SRPP designs omit these resistors.

I don't think they have some real purpose, but can't say for sure.

The high voltage shunt regulators / constant current sources are the second group of items seen here. One for each channel will go between C301 and R306 in the Clarinet schematic. Coincidentally, if you visit the link and click on "Power Amplifier" on the left menu, you will see the same tube amplifier I am now using and plan to use in conjunction with the Clarinet.

While generally I also prefer to use some sort of  advanced power supply design - I don't think in this case a CCS would be of any good. Shunt regulators? Maybe yes, especially with SS rectifiers.

BTW, your list of Russian tubes was very interesting. I was particularly curious about employing a pair of 6Н6П tubes in a circuit like the Clarinet...

6Н6П is too "heavy" for the circuit IMHO. I'd better use 6Н1П.

[Brinkman]

I don't think they have some real purpose, but can't say for sure.

Just asked Jim about those 220 ohm resistors. They're plate stoppers for RF oscillations.

I'd better use 6Н1П.

I printed out the datasheets today. Looks good! When can we expect a 6Н1П Clarinet (Clar1Пet?) circuit Poty? 

[poty]

All is easy in this world! (about 220 Ohm resistors)
What do you mean about 6Н1П SRPP? I use it already for about a year. My own PCBs, which allowed to use any tube (Russian, 12AU7, octal)... I use also different PCBs for power supply (also mine). The PS PCBs not very easy to build, because they are also "universal" (can be used as tube-based or SS-based PS, RCRC filtering, LCLC filtering, regulated or not, low-noise...). I build 3 devices with them (integrated amplifiers) and understand that probably it is too difficult to use such PCBs. In the future I'll plan to divide the PS PCB into "standard size" blocks which could be combined to achieve the necessary level of filtering and regulating.
But... It's a pity, but the idea is not mine. So I can't add my name to the resulting device.